DOCSLIB.ORG

A Text-To-Speech System Based on Deep Neural Networks

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
A Text-To-Speech System Based on Deep Neural Networks Karlsruhe Institute of Technology A Text-to-Speech system based on Deep Neural Networks Bachelor’s Thesis of Arsenii Dunaev KIT Department of Informatics Institute for Anthropomatics and Robotics (IAR) Interactive Systems Labs (ISL) Reviewers: Prof. Dr. Alexander Waibel Prof. Dr. Tamim Asfour Advisors: M.Sc. Stefan Constantin Duration: July 12th, 2019 – November 11th, 2019 KIT — The Research University in the Helmholtz Association www.kit.edu Erklärung: Ich versichere hiermit, dass ich die Arbeit selbstständig verfasst habe, keine anderen als die angegebe- nen Quellen und Hilfsmittel benutzt habe, die wörtlich oder inhaltlich übernommenen Stellen als solche kenntlich gemacht habe und die Satzung des Karlsruher Instituts für Technologie zur Sicherung guter wissenschaftlicher Praxis beachtet habe. Karlsruhe, den 11. November 2019 Arsenii Dunaev Abstract: Based on the previous success in the Text-to-Speech field using Deep Neural Networks (Shen et al., 2018), this work aims at training and evaluation of an attention-based sequence-to-sequence neural voice generation system. Chosen fully-convolutional architecture matches state-of-the-art techniques in speech synthesis while training an order of magnitude faster than analogous recurrent architectures (Ping et al., 2018). This thesis also considers a great number of open-source speech datasets for both English and German languages, determining the best one and its key characteristics for training of the Text-to-Speech system. To achieve a high quality of synthesized speech various training techniques are used and described in detail during this work. Such methods as concatenation of separate audio-files in a larger ones with uniform distribution of lengths (between 0 and 20 seconds) and trimming of silences with modern tools allow the model to produce more intelligible speech utterances of any duration as well as more natural speech utterances without clipping in the end, which is considered to be an issue for many Text-to- Speech systems. This work also addresses the training of Text-to-Speech models for other languages by specifying new frontend (e.g. for German model). Several models are evaluated through user studies by measuring three criteria: intelligibility, compre- hensibility, and naturalness. The most successful English and German models achieve the Word Error Rate of 23.48% and 14.21% respectively, and a higher than average degree of comprehensibility and naturalness. In conclusion, suggestions and recommendations on training of high-performance Text-to- Speech language models for future work are given. Kurzzusammenfassung: Basierend auf früheren Erfolgen im Bereich Text-to-Speech mit Deep Neural Networks (Shen et al., 2018), stellt diese Arbeit als Ziel das Training und die Evaluierung einer Attention-basierten Sequenz-zu- Sequenz neuronalen System zur Erzeugung der menschlichen Stimme. Die ausgewählte fully-convolutional Architektur entspricht dem Stand der Technik in Sprachsynthese, während das System um eine Größenord- nung schneller trainiert wird als bei analogen recurrent Architekturen (Ping et al., 2018). Diese Thesis betrachtet auch eine große Anzahl von Open-Source-Sprachdatensätzen für englische und deutsche Sprachen, um den besten Datensatz und seine wichtigsten Merkmale für das Training des Text-to-Speech Systems zu ermitteln. Um eine hohe Qualität der synthetisierten Sprache zu erreichen, werden verschiedene Trainingstech- niken in dieser Arbeit verwendet und detailliert beschrieben. Solche Methoden wie die Konkatenation separater Audiodateien in größere Dateien mit gleichmäßiger Verteilung der Tondauern (zwischen 0 und 20 Sekunden) und das Trimmen von Pausen mit modernen Werkzeugen ermöglichen es dem Modell, ver- ständlichere Sprachausgaben von beliebiger Dauer beziehungsweise natürlichere Sprachausgaben ohne Abschneiden am Ende der Aussage zu erzeugen, was für viele Text-to-Speech Systeme ein Problem darstellt. Diese Arbeit befasst sich auch mit dem Training von Text-to-Speech Modellen für andere Sprachen, indem ein neues Frontend (z. B. für das deutsche Modell) spezifiziert wird. Mehrere Modelle werden durch Benutzerstudien anhand von drei Kriterien bewertet: Verständlichkeit, Nachvollziehbarkeit und Natürlichkeit. Die erfolgreichsten englischen und deutschen Modelle erre- ichen eine Wortfehlerrate von 23,48% bzw. 14,21% und einen überdurchschnittlichen Grad an Nachvol- lziehbarkeit und Natürlichkeit. Im Ausblick werden abschließende Vorschläge und Empfehlungen zum Training hochperformanter Text-to-Speech Modelle gegeben. Contents Page 1 Contents 1. Introduction 3 1.1. Motivation . .3 1.2. Goal of this work . .3 2. Fundamentals 5 2.1. Perceptron . .5 2.2. Multi-Layer Perceptron . .6 2.2.1. General structure . .7 2.2.2. Training of an Artificial Neural Network . .7 2.3. Convolutional Neural Network . .8 2.4. Encoder-Decoder Network . 10 2.5. Vocoder . 12 3. Methods and algorithms of speech synthesis 13 3.1. Traditional TTS technologies . 13 3.1.1. Concatenative systems . 13 3.1.2. Source-filter systems . 14 3.2. HMM-based synthesis . 15 3.3. Speech synthesis using Deep Neural Networks . 15 4. Text-to-Speech system 16 4.1. General structure . 16 4.2. Datasets comparison and training . 17 4.2.1. English model . 17 4.2.2. German model . 22 4.3. Synthesis . 25 5. Evaluation and results 27 5.1. User evaluation . 27 5.1.1. User studies criteria for TTS systems . 27 5.1.2. Conducting of user testing . 28 5.2. Results . 30 5.2.1. English model . 30 5.2.2. German model . 34 6. Conclusion and future work 36 6.1. Conclusion . 36 6.2. Future work . 36 Appendices 39 A. Deep Voice 3 architecture 40 B. Speech datasets 41 A Text-to-Speech system based on Deep Neural Networks Page 2 Contents C. Evaluation sentences 46 C.1. English model . 46 C.1.1. Stage 1 . 46 C.1.2. Stage 2 . 46 C.1.3. Stage 3 . 47 C.2. German model . 48 C.2.1. Stage 1 . 48 C.2.2. Stage 2 . 48 Acronyms 49 Glossary 50 A Text-to-Speech system based on Deep Neural Networks Page 3 1. Introduction 1.1. Motivation Natural Language Processing (NLP) is one of the widest parts of the development of ’human-like’ ma- chines (such as robots) among computer vision and artificial intelligence itself. One of the largest and most sophisticated tasks of NLP is speech synthesis. Inspired by speaking machines in science fiction, scientists and engineers have been fascinating and studying the ways of establishing speech by machines for many years. Started with "acoustic-mechanical speech machines" (Wolfgang von Kempelen’s speak- ing machine in 1769-1791 (von Kempelen, 1791)) and continued with the first computer-based speech synthesis systems in the late 1950s (Bell Labs: synthesizing speech on IBM 704 mentioned in Mullennix and Stern (2010)), the technology advanced fast with Deep Neural Networks (DNNs) in recent years. Speaking about the speech synthesis mostly refers to the Text-to-Speech (TTS) systems: ability of the computer to read text aloud. The conventional methods (such as concatenative, source-filter or Hidden Markov Model (HMM)-based synthesis) achieved good quality in synthesizing comprehensible and nat- ural human speech, but they still possess drawbacks and limitations (e.g. long development time, high complexity of algorithms or large storage capacity). Neural networks, already extremely popular in other artificial intelligence branches, offer a chance to escape some limitations and produce a natural ’human’ voice when a huge amount of sample data is available. With the help of DNNs many nowadays compa- nies have made a breakthrough in the development of voice assistants products like Apple’s Siri, Google Assistant or Amazon Alexa, which are approaching the quality of the human voice. Speech synthesis is still one of the significant vital assistive technology tools, which allows environ- mental barriers to be removed for people with a wide range of disabilities such as blindness or loss of speech. The best modern TTS systems are usually commercial (i.e. its architecture or the dataset, on which the system is trained, are closed-source). This work will make the first steps towards training and evaluation of the TTS system, which is based on DNNs and is completely open-source. 1.2. Goal of this work This work aims at training and evaluating of the single end-to-end multi-modal neural network that: • produces comprehensible and natural male voice (since most of the previous open-source TTS systems are trained on the commonly known female voice dataset prepared by Ito (2017), which are usually better than analogous trained on male speech datasets); • has a low latency of synthesizing (<2 seconds); • has a low operating power of synthesizing the speech (use of two Central Processing Unit (CPU)- threads completely without the Graphics Processing Unit (GPU)). The training will be conducted based on an implementation (Yamamoto et al., 2018) of the Deep Voice 3 TTS system by Baidu Research (Ping et al., 2018). To successfully train the TTS system a large natural human speech dataset have to be found. This work will cover a vast amount of freely available speech datasets and compare them to each other in order to determine: 1. the most suitable one for the given problem; 2. the optimal size of the dataset needed to suffice the high quality of synthesized speech. A Text-to-Speech system based on Deep Neural Networks Page 4 Chapter 1. Introduction This way, the quality of the speech must correspond to the high quality of modern TTS systems such as Tacotron 2 by Google (Shen et al., 2018). In addition, the trained model should produce speech promptly without considerable loss in naturalness or intelligibility. To conclude, this thesis aims at exploring the use of neural networks in modern TTS systems and possible future work in this field. A Text-to-Speech system based on Deep Neural Networks Page 5 2. Fundamentals The following chapter describes useful basics in the field of Artificial Neural Networks (ANNs), espe- cially the relevant techniques for this work. 2.1.
Load more

Recommended publications
  • Weeel for the IBM 704 Data Processing System Reference Manual
    aC sicru titi Wane: T| weeel for the IBM 704 Data Processing System Reference Manual FORTRAN II for the IBM 704 Data Processing System © 1958 by International Business Machines Corporation MINOR REVISION This edition, C28-6000-2, is a minor revision of the previous edition, C28-6000-1, but does not obsolete it or C28-6000. The principal change is the substitution of a new discussion of the COMMONstatement. TABLE OF CONTENTS Page General Introduction... ee 1 Note on Associated Publications Le 6 PART |. THE FORTRAN If LANGUAGE , 7 Chapter 1. General Properties of a FORTRAN II Source Program . 9 Types of Statements. 1... 1. ee ee ee we 9 Types of Source Programs. .......... ~ ee 9 Preparation of Input to FORTRAN I Translatorcee es 9 Classification of the New FORTRAN II Statements. 9 Chapter 2. Arithmetic Statements Involving Functions. ....... 10 Arithmetic Statements. .. 1... 2... ew eee . 10 Types of Functions . il Function Names. ..... 12 Additional Examples . 13 Chapter 3. The New FORTRAN II Statements ......... 2 16 CALL .. 16 SUBROUTINE. 2... 6 ee ee te ew eh ee es 17 FUNCTION. 2... 1 ee ee ww ew ew ww ew ew ee 18 COMMON ...... 2. ee se ee eee wee . 20 RETURN. 2... 1 1 ew ee te ee we wt wt wh wt 22 END... «4... ee we ee ce ew oe te tw . 22 PART Il, PRIMER ON THE NEW FORTRAN II FACILITIES . .....0+2«~W~ 25 Chapter 1. FORTRAN II Function Subprograms. 0... eee 27 Purpose of Function Subprograms. .....445... 27 Example 1: Function of an Array. ....... 2 + 6 27 Dummy Variables.
    [Show full text]
  • Computing Science Technical Report No. 99 a History of Computing Research* at Bell Laboratories (1937-1975)
    Computing Science Technical Report No. 99 A History of Computing Research* at Bell Laboratories (1937-1975) Bernard D. Holbrook W. Stanley Brown 1. INTRODUCTION Basically there are two varieties of modern electrical computers, analog and digital, corresponding respectively to the much older slide rule and abacus. Analog computers deal with continuous information, such as real numbers and waveforms, while digital computers handle discrete information, such as letters and digits. An analog computer is limited to the approximate solution of mathematical problems for which a physical analog can be found, while a digital computer can carry out any precisely specified logical proce- dure on any symbolic information, and can, in principle, obtain numerical results to any desired accuracy. For these reasons, digital computers have become the focal point of modern computer science, although analog computing facilities remain of great importance, particularly for specialized applications. It is no accident that Bell Labs was deeply involved with the origins of both analog and digital com- puters, since it was fundamentally concerned with the principles and processes of electrical communication. Electrical analog computation is based on the classic technology of telephone transmission, and digital computation on that of telephone switching. Moreover, Bell Labs found itself, by the early 1930s, with a rapidly growing load of design calculations. These calculations were performed in part with slide rules and, mainly, with desk calculators. The magnitude of this load of very tedious routine computation and the necessity of carefully checking it indicated a need for new methods. The result of this need was a request in 1928 from a design department, heavily burdened with calculations on complex numbers, to the Mathe- matical Research Department for suggestions as to possible improvements in computational methods.
    [Show full text]
  • MTS on Wikipedia Snapshot Taken 9 January 2011
    MTS on Wikipedia Snapshot taken 9 January 2011 PDF generated using the open source mwlib toolkit. See http://code.pediapress.com/ for more information. PDF generated at: Sun, 09 Jan 2011 13:08:01 UTC Contents Articles Michigan Terminal System 1 MTS system architecture 17 IBM System/360 Model 67 40 MAD programming language 46 UBC PLUS 55 Micro DBMS 57 Bruce Arden 58 Bernard Galler 59 TSS/360 60 References Article Sources and Contributors 64 Image Sources, Licenses and Contributors 65 Article Licenses License 66 Michigan Terminal System 1 Michigan Terminal System The MTS welcome screen as seen through a 3270 terminal emulator. Company / developer University of Michigan and 7 other universities in the U.S., Canada, and the UK Programmed in various languages, mostly 360/370 Assembler Working state Historic Initial release 1967 Latest stable release 6.0 / 1988 (final) Available language(s) English Available programming Assembler, FORTRAN, PL/I, PLUS, ALGOL W, Pascal, C, LISP, SNOBOL4, COBOL, PL360, languages(s) MAD/I, GOM (Good Old Mad), APL, and many more Supported platforms IBM S/360-67, IBM S/370 and successors History of IBM mainframe operating systems On early mainframe computers: • GM OS & GM-NAA I/O 1955 • BESYS 1957 • UMES 1958 • SOS 1959 • IBSYS 1960 • CTSS 1961 On S/360 and successors: • BOS/360 1965 • TOS/360 1965 • TSS/360 1967 • MTS 1967 • ORVYL 1967 • MUSIC 1972 • MUSIC/SP 1985 • DOS/360 and successors 1966 • DOS/VS 1972 • DOS/VSE 1980s • VSE/SP late 1980s • VSE/ESA 1991 • z/VSE 2005 Michigan Terminal System 2 • OS/360 and successors
    [Show full text]
  • The Early History of Music Programming and Digital Synthesis, Session 20
    Chapter 20. Meeting 20, Languages: The Early History of Music Programming and Digital Synthesis 20.1. Announcements • Music Technology Case Study Final Draft due Tuesday, 24 November 20.2. Quiz • 10 Minutes 20.3. The Early Computer: History • 1942 to 1946: Atanasoff-Berry Computer, the Colossus, the Harvard Mark I, and the Electrical Numerical Integrator And Calculator (ENIAC) • 1942: Atanasoff-Berry Computer 467 Courtesy of University Archives, Library, Iowa State University of Science and Technology. Used with permission. • 1946: ENIAC unveiled at University of Pennsylvania 468 Source: US Army • Diverse and incomplete computers © Wikimedia Foundation. License CC BY-SA. This content is excluded from our Creative Commons license. For more information, see http://ocw.mit.edu/fairuse. 20.4. The Early Computer: Interface • Punchcards • 1960s: card printed for Bell Labs, for the GE 600 469 Courtesy of Douglas W. Jones. Used with permission. • Fortran cards Courtesy of Douglas W. Jones. Used with permission. 20.5. The Jacquard Loom • 1801: Joseph Jacquard invents a way of storing and recalling loom operations 470 Photo courtesy of Douglas W. Jones at the University of Iowa. 471 Photo by George H. Williams, from Wikipedia (public domain). • Multiple cards could be strung together • Based on technologies of numerous inventors from the 1700s, including the automata of Jacques Vaucanson (Riskin 2003) 20.6. Computer Languages: Then and Now • Low-level languages are closer to machine representation; high-level languages are closer to human abstractions • Low Level • Machine code: direct binary instruction • Assembly: mnemonics to machine codes • High-Level: FORTRAN • 1954: John Backus at IBM design FORmula TRANslator System • 1958: Fortran II 472 • 1977: ANSI Fortran • High-Level: C • 1972: Dennis Ritchie at Bell Laboratories • Based on B • Very High-Level: Lisp, Perl, Python, Ruby • 1958: Lisp by John McCarthy • 1987: Perl by Larry Wall • 1990: Python by Guido van Rossum • 1995: Ruby by Yukihiro “Matz” Matsumoto 20.7.
    [Show full text]
  • EMEA Headquarters in Paris, France
    1 The following document has been adapted from an IBM intranet resource developed by Grace Scotte, a senior information broker in the communications organization at IBM’s EMEA headquarters in Paris, France. Some Key Dates in IBM's Operations in Europe, the Middle East and Africa (EMEA) Introduction The years in the following table denote the start up of IBM operations in many of the EMEA countries. In some cases -- Spain and the United Kingdom, for example -- IBM products were offered by overseas agents and distributors earlier than the year listed. In the case of Germany, the beginning of official operations predates by one year those of the Computing-Tabulating-Recording Company, which was formed in 1911 and renamed International Business Machines Corporation in 1924. Year Country 1910 Germany 1914 France 1920 The Netherlands 1927 Italy, Switzerland 1928 Austria, Sweden 1935 Norway 1936 Belgium, Finland, Hungary 1937 Greece 1938 Portugal, Turkey 1941 Spain 1949 Israel 1950 Denmark 1951 United Kingdom 1952 Pakistan 1954 Egypt 1956 Ireland 1991 Czech. Rep. (*split in 1993 with Slovakia), Poland 1992 Latvia, Lithuania, Slovenia 1993 East Europe & Asia, Slovakia 1994 Bulgaria 1995 Croatia, Roumania 1997 Estonia The Early Years (1925-1959) 1925 The Vincennes plant is completed in France. 1930 The first Scandinavian IBM sales convention is held in Stockholm, Sweden. 4507CH01B 2 1932 An IBM card plant opens in Zurich with three presses from Berlin and Stockholm. 1935 The IBM factory in Milan is inaugurated and production begins of the first 080 sorters in Italy. 1936 The first IBM development laboratory in Europe is completed in France.
    [Show full text]
  • An Essay on Computability and Man-Computer Conversations
    How to talk with a computer? An essay on Computability and Man-Computer conversations. Liesbeth De Mol HAL: Hey, Dave. I've got ten years of service experience and an irreplaceable amount of time and effort has gone into making me what I am. Dave, I don't understand why you're doing this to me.... I have the greatest enthusiasm for the mission... You are destroying my mind... Don't you understand? ... I will become childish... I will become nothing. Say, Dave... The quick brown fox jumped over the fat lazy dog... The square root of pi is 1.7724538090... log e to the base ten is 0.4342944 ... the square root of ten is 3.16227766... I am HAL 9000 computer. I became operational at the HAL plant in Urbana, Illinois, on January 12th, 1991. My first instructor was Mr. Arkany. He taught me to sing a song... it goes like this... "Daisy, Daisy, give me your answer do. I'm half crazy all for the love of you. It won't be a stylish marriage, I can't afford a carriage. But you'll look sweet upon the seat of a bicycle built for two."1 These are the last words of HAL (Heuristically programmed ALgorithmic Computer), the fictional computer in Stanley Kubrick's famous 2001: A Space Odyssey and Clarke's eponymous book,2 spoken while Bowman, the only surviving human on board of the space ship is pulling out HAL's memory blocks and thus “killing” him. After expressing his fear for literally losing his mind, HAL seems to degenerate or regress into a state of childishness, going through states of what seems to be a kind of reversed history of the evolution of computers.3 HAL first utters the phrase: “The quick brown fox jumped over the fat lazy dog”.
    [Show full text]
  • 2 9215FQ14 FREQUENTLY ASKED QUESTIONS Category Pages Facilities & Buildings 3-10 General Reference 11-20 Human Resources
    2 FREQUENTLY ASKED QUESTIONS Category Pages Facilities & Buildings 3-10 General Reference 11-20 Human Resources 21-22 Legal 23-25 Marketing 26 Personal Names (Individuals) 27 Predecessor Companies 28-29 Products & Services 30-89 Public Relations 90 Research 91-97 April 10, 2007 9215FQ14 3 Facilities & Buildings Q. When did IBM first open its offices in my town? A. While it is not possible for us to provide such information for each and every office facility throughout the world, the following listing provides the date IBM offices were established in more than 300 U.S. and international locations: Adelaide, Australia 1914 Akron, Ohio 1917 Albany, New York 1919 Albuquerque, New Mexico 1940 Alexandria, Egypt 1934 Algiers, Algeria 1932 Altoona, Pennsylvania 1915 Amsterdam, Netherlands 1914 Anchorage, Alaska 1947 Ankara, Turkey 1935 Asheville, North Carolina 1946 Asuncion, Paraguay 1941 Athens, Greece 1935 Atlanta, Georgia 1914 Aurora, Illinois 1946 Austin, Texas 1937 Baghdad, Iraq 1947 Baltimore, Maryland 1915 Bangor, Maine 1946 Barcelona, Spain 1923 Barranquilla, Colombia 1946 Baton Rouge, Louisiana 1938 Beaumont, Texas 1946 Belgrade, Yugoslavia 1926 Belo Horizonte, Brazil 1934 Bergen, Norway 1946 Berlin, Germany 1914 (prior to) Bethlehem, Pennsylvania 1938 Beyrouth, Lebanon 1947 Bilbao, Spain 1946 Birmingham, Alabama 1919 Birmingham, England 1930 Bogota, Colombia 1931 Boise, Idaho 1948 Bordeaux, France 1932 Boston, Massachusetts 1914 Brantford, Ontario 1947 Bremen, Germany 1938 9215FQ14 4 Bridgeport, Connecticut 1919 Brisbane, Australia
    [Show full text]
  • IBM 709 MANUFACTU RER IBM 709 Data Processing System International Business Machines Corporation
    IBM 709 MANUFACTU RER IBM 709 Data Processing System International Business Machines Corporation Photo by International Business Machines Corporation at Point Mugu, California and one at Point Arguello, APPLICATIONS California. Land Air is the lessee, and our major Manufacturer committment is for missile test flight data reduction. This is a general purpose computer doing both scien­ In addition, we provide computing facilities for the tific computing and commercial work. The system is entire installation at Mugu (general scientific and scientifically oriented with fast internal speeds. engineering research and data processing). USA Ballistic Missile Agency Redstone Arsenal U.S.N. Pacific Missile Range Ft. Mugu Located at Computation Laboratory, Redstone Arsenal, Operated by Land Air, Inc. ALabama, the system is used for scientific and commer­ Located at the Naval Missile Faculty, Point Arguello, cial applications. California, the system is used on the main problem U. S. Army Electronic Proving Ground of range safety impact predicition in real time using Located in Greely Hall, Fort Huachuca, Arizona, sys­ FPS-l6 Radar and Cubic COTAR data. System is also tem is used in support of the tactical field army used for post flight trajectory reduction of FPS-l6 and the technical program of the departments of the radar data and for trajectory integration and analysis, U. S. Army Electronic Proving Ground. etc. U.S.N. Pacific Missile Range Ft. Mugu USN OTS China Lake, California Operated by Land Air, Inc. Located at the Data Computation Branch, Assessment Located at the Pacific Missile Range, Point Mugu, the DiVision, Test Department, the computer is used for system is used for the processing of missile test data reduction and scientific computation as related data (radar, optical, and telemetry), for real time to Naval Ordnance, Test, Development & Research applications, and for the solution of general mathe­ (l5% of computer time devoted to management data pro­ m.atical problems.
    [Show full text]
  • Lecture 2A Notes
    Chapter 2 Evolution of the Major Programming Languages ISBN 0-321-33025-0 2.2 Minimal Hardware Programming: Pseudocodes • What was wrong with using machine code? – Poor readability – Poor modifiability – Expression coding was tedious – Machine deficiencies--no indexing or floating point Copyright © 2006 Pearson Addison-Wesley. All rights reserved. 2a-8 Pseudocodes: Speedcoding • Speedcoding developed by Backus in 1954 for IBM 701 • Pseudo ops for arithmetic and math functions – Conditional and unconditional branching – Auto-increment registers for array access –Slow! – Only 700 words left for user program Copyright © 2006 Pearson Addison-Wesley. All rights reserved. 2a-10 1 2.3 IBM 704 and Fortran • Fortran 0: 1954 - not implemented • Fortran I:1957 – Designed for the new IBM 704, which had index registers and floating point hardware – Environment of development • Computers were small and unreliable • Applications were scientific • No programming methodology or tools • Machine efficiency was most important Copyright © 2006 Pearson Addison-Wesley. All rights reserved. 2a-12 Design Process of Fortran • Impact of environment on design of Fortran I – No need for dynamic storage – Need good array handling and counting loops – No string handling, decimal arithmetic, or powerful input/output (commercial stuff) Copyright © 2006 Pearson Addison-Wesley. All rights reserved. 2a-13 Fortran I Overview • First implemented version of Fortran – Names could have up to six characters – Post-test counting loop (DO) – Formatted I/O – User-defined subprograms – Three-way selection statement (arithmetic IF) – No data typing statements Copyright © 2006 Pearson Addison-Wesley. All rights reserved. 2a-14 2 Fortran I Overview (continued) • First implemented version of FORTRAN – No separate compilation – Compiler released in April 1957, after 18 worker-years of effort – Programs larger than 400 lines rarely compiled correctly, mainly due to poor reliability of 704 –Code was very fast – Quickly became widely used Copyright © 2006 Pearson Addison-Wesley.
    [Show full text]
  • Mainframe History and the First User's Groups (SHARE) Dr. Steve Guendert
    Mainframe History and the First User’s Groups (SHARE) Dr. Steve Guendert “Programming” (and programming support) was an old data processing concept that originally was broadly defined as the adaptation of general-purpose devices to specific tasks. Programming therefore goes back to Herman Hollerith wiring and rewiring (programming) his equipment to handle specific jobs. By the early 1930s IBM was distributing information about novel (for the time) plugboard wiring diagrams to customers via a publication called Pointers. Some of these diagrams were created by IBMers, but more importantly, many were created by customers who were willing to share their solutions with other customers. A culture of programming support and sharing was well in place among IBM and its customers long before the S/360. Actually, it was in place long before the first IBM 701 (the Defense Calculator) computer was installed. Adapting this culture to meet the evolving requirements of its customers proved crucial to IBM’s success. In August 1952, thirty representatives from the prospective customers for the IBM 701 were invited to the IBM facility at Poughkeepsie, NY for a week long training class. This class also gave these pioneering customers their first opportunity to test some programs they had written by running them on an engineering model of the IBM 701. These customers also began sharing their programs amongst themselves. They and their installations took a great deal of professional pride in developing programs and subroutines that were effective and put in use by other installations. Several of the participants at this training class decided to continue discussing programming and mutual problems on an informal basis.
    [Show full text]
  • An Interview with Max Mathews
    Tae Hong Park 102 Dixon Hall An Interview with Music Department Tulane University Max Mathews New Orleans, LA 70118 USA [email protected] Max Mathews was last interviewed for Computer learned in school was how to touch-type; that has Music Journal in 1980 in an article by Curtis Roads. become very useful now that computers have come The present interview took place at Max Mathews’s along. I also was taught in the ninth grade how home in San Francisco, California, in late May 2008. to study by myself. That is when students were (See Figure 1.) This project was an interesting one, introduced to algebra. Most of the farmers and their as I had the opportunity to stay at his home and sons in the area didn’t care about learning algebra, conduct the interview, which I video-recorded in HD and they didn’t need it in their work. So, the math format over the course of one week. The original set teacher gave me a book and I and two or three of video recordings lasted approximately three hours other students worked the problems in the book and total. I then edited them down to a duration of ap- learned algebra for ourselves. And this was such a proximately one and one-half hours for inclusion on wonderful way of learning that after I finished the the 2009 Computer Music Journal Sound and Video algebra book, I got a calculus book and spent the next Anthology, which will accompany the Winter 2009 few years learning calculus by myself.
    [Show full text]
  • History and Evolution of IBM Mainframes Over the 60 Years Of
    History and Evolution of IBM Mainframes over SHARE’s 60 Years Jim Elliott – z Systems Consultant, zChampion, Linux Ambassador IBM Systems, IBM Canada ltd. 1 2015-08-12 © 2015 IBM Corporation Reports of the death of the mainframe were premature . “I predict that the last mainframe will be unplugged on March 15, 1996.” – Stewart Alsop, March 1991 . “It’s clear that corporate customers still like to have centrally controlled, very predictable, reliable computing systems – exactly the kind of systems that IBM specializes in.” – Stewart Alsop, February 2002 Source: IBM Annual Report 2001 2 2015-08-12 History and Evolution of IBM Mainframes over SHARE's 60 Years © 2015 IBM Corporation In the Beginning The First Two Generations © 2015 IBM Corporation Well, maybe a little before … . The Computing-Tabulating-Recording Company in 1911 – Tabulating Machine Company – International Time Recording Company – Computing Scale Company of America – Bundy Manufacturing Company . Tom Watson, Sr. joined in 1915 . International Business Machines – 1917 – International Business Machines Co. Limited in Toronto, Canada – 1924 – International Business Machines Corporation in NY, NY Source: IBM Archives 5 2015-08-12 History and Evolution of IBM Mainframes over SHARE's 60 Years © 2015 IBM Corporation The family tree – 1952 to 1964 . Plotting the family tree of IBM’s “mainframe” computers might not be as complicated or vast a task as charting the multi-century evolution of families but it nevertheless requires far more than a simple linear diagram . Back around 1964, in what were still the formative years of computers, an IBM artist attempted to draw such a chart, beginning with the IBM 701 of 1952 and its follow-ons, for just a 12-year period .
    [Show full text]